1. Field of the Invention
Exemplary embodiments of the present invention relate to a method and apparatus for measuring, based on a composition for automatic image correction, the exposure degree of an object included in an image to automatically correct the image.
2. Discussion of the Background
When human eyes observe a visual image, unlike a camera, the eyes tend to focus on a particular portion or object of the entire image. That is, a human visual cognitive system tends to repeat focusing on and recognizing one object or portion at a sight and accumulate the recognized objects or portions to collectively constitute one image in the brain of the observer.
On the other hand, a camera uses a single lens. Therefore, to render an image taken by a camera in a similar manner to the human recognition system, the taken image needs to be divided into multiple portions or objects and then integrated when being corrected, rather than being corrected with respect to the entire image.
However, because the exposure degree of the pixels constituting an image is conventionally measured with respect to the entire region of the image for image correction, accuracy in measuring the exposure degree with respect to the object is not satisfactory. Generally, exposure refers to transmission of light that falls on a camera film where an image is formed. An exposure degree (or amount) refers to the level of brightness of the image, determined by transmission of light.
It is believed that, for more accurate image correction, the exposure degree of a main object included in an image needs to be measured. However, the process for finding the main object in the image demands high central processing unit (CPU) cost. Furthermore, when finding the object, even a region other than the object may erroneously be recognized as the object.
For example, according to a method for detecting an object by detecting high-frequency components, which are regions with a high change rate, as disclosed in KR Patent Application Publication No. 10-2011-0126616 (published on Nov. 23, 2011 and entitled “Image Processing Device and Method”), not only an object 110 but also a region 120 including high frequency components, such as the background, may be detected as an object as shown in FIGS. 1A and 1B. In this case, image correction may not be accurately performed.